Role of fluid properties in gas transfer - ACS Publications

Oak Ridge Gaseous Diffusion Plant, operated by Union Car- ... H. Then for streams. dCL .... 0 —. (7). In Equation 7 0is the surfacetension of a clea...
0 downloads 0 Views 883KB Size
Literature Cited

Burnet, F. M., Stanley, W. M., eds, “The Viruses,” Vol. 1, New York, Academic Press, 1959. Lammers, W. T., Limnol. Oceanogr., in press (1966a). Lammers, W. T., Verhandl. Intern. Verein. Limnol. 15, 1021 (1964). Lammers, W. T., Verhandl. Intern. Verein. Limnol. 16, in press (1966b).

Smith, K. M., Lauffer, M. A., Eds., “Advances in Virus Research,” Vol. 9, Academic Press, New York, 1962. Received for review October 14, 1966. Accepted December 1 , 1966. Division of Water, Air, and Waste Chemistry, l52nd Meeting. ACS, New York, N . Y., September 1966. Work performed at the Dauidson College, Davidson, N . C., and the Oak Ridge Gaseous Digusion Plant, operated by Union Carbide Corp. for the United States Atomic Energy Cornniission.

Role of Fluid Properties in Gas Transfer Ivan Metzger, Newark College of Engineering, Newark, N. J., and William E. Dobbins, New York University, Bronx, N. Y.

Considerable attention has been directed to the determination of the capacity of a river to assimilate organic pollution. One aspect of this problem is the prediction of river aeration, which depends on an adequate knowledge of the mechanism of absorption of low-solubility gases into a turbulent fluid. Reported herein are the results of an experimental and theoretical study which was undertaken to extend the usefulness of previous work on the film penetration model for gas absorption coefficients. Proposed relationships for surface renewal frequency and the film thickness were verified and then used in the model to predict effects of temperature on oxygen absorption which are in good agreement with values reported in the literature.

T

he rate of change of concentration of dissolved gas in a liquid of volume, V , which is assumed to be of uniform concentration, CL,is dCL _ -- K L A ~ C-, CL) dt V

(1)

where C, is the saturation concentration. A, is the surface area, and Kr, is the absorption coefficient. Solution of Equation 1 for an initial liquid concentration of C, yields the familiar logarithmic absorption equation

The actual surface area, A,, is difficult to measure in turbulent fluids and the horizontal projection of the actual surface area, A,, is often used. For this discussion, they are related by A,

=

CAAO

(3)

where Ca is a coefficient depending on flow conditions and other factors. Absorption coefficients computed by using A, are designated by primes in this work. They are related to those calculated by using A , by KL’

=

CAKL

(4)

where KL ’ is termed the apparent absorption coefficient and KL is termed the true absorption coefficient. When these equations are applied to streams it is common to

refer to the quantity (C, - CL) as D, the dissolved oxygen deficit. Further, the quantity A,/V equals the average depth, H. Then for streams dCL - KL’A,D - KL’D dt V H

=

K2D

where K2is termed the reaeration coefficient, which is equal to the apparent absorption coefficient divided by the depth. Models for Gas Absorption

Various theories of gas absorption result in different equations for the determination of the absorption coefficient, KL. Lewis and Whitman (1924) assumed that there is a stagnant film at the liquid surface in which steady-state molecular transfer controls the rate of mass transfer. This theory results in a value of KL equal to the molecular diffusivity, D,, divided by the film thickness, L. Conversely, the penetration theories of Higbie (1935) and Danckwerts (1951) abandon the film concept and propose that the interface is continuously replaced by eddies which control the rate by unsteady-state molecular transfer, Their models result in values of KL proportional to the square root of D,. The divergence of these results concerning KL may have been resolved with the suggestion that the film may be assumed to maintain its existence in a statistical sense (Dobbins, 1956). The film is considered to be always present; but its composition is continuously changed by liquid from beneath the surface. Indeed, surface films have been reported (Bolin, 1960; Kanwisher, 1963; Davis and Crandall, 1930), and the proposed stochastic nature of the existence of the film is in accord with ideas of fluid turbulence. The resulting film penetration model for KL is KL

=

dD,rcoth

$E

(6)

where r is the frequency of replacement of the liquid film. This equation reduces to Danckwerts’ equation ( K L = as the coth term becomes about 3 or greater and approaches the film equation ( K L = D,/L) as the renewal rate, r, approaches zero. The effect of molecular diffusivity on the absorption coefficient has often been measured in attempts to verify various theories of gas absorption. This effect is usually expressed by showing how the value of KL varies with D, raised to an

dG)

Volume 1, Number 1, January 1967 57

exponent, n. The Lewis and Whitman theory is represented by an exponent of unity, while the theories of Higbie and Danckwerts have values of 0.5. It is significant that Equation 6 has the property that the exponent can vary from 0.5 to unity. Experimental data previously reported (Dobbins, 1964) demonstrated a variation in the exponent for a n aeration system with the turbulence generated from below the surface. The higher values of the exponent were obtained for experiments conducted at low mixing intensities. Subsequent experiments in a similar system (Metzger, 1965) showed that n approaches a constant value at high mixing intensities. This constant appears to be 0.5, in accord with the predictions of Equation 6, although the exact value depends on a precise knowledge of the molecular diffusion coefficients. Equation 6 was further verified by direct application to stream aeration (Dobbins, 1964). Absorption coefficients reported for a wide range of stream conditions were compared with those computed using Equation 6 along with those computed using the O’Connor-Dobbins equations (1956) and the Churchill-Elmore-Buckingham formula (1962). This latter comparison demonstrated that Equation 6 provided results superior to the other formulas over a broad spectrum of conditions. The film penetration model may be employed as a workable representation of the gas absorption phenomenon as it is encountered in natural rivers. However, its usefulness is somewhat limited by the lack of a verified theory for the determination of r, the frequency of replacement of the liquid film, and L, the film thickness. Proposed Equations ,for r and L

The surface of pure water is not a monomolecular layer with unaltered liquid immediately underneath it, but a zone in which molecules are oriented (Drost-Hansen, 1965). The orientation is in the direction of polarization (Weyl, 1951) and may extend to depths of 10-5 cm. (Henniker, 1949). It is to this orientation that the surface tension and elastic strength (Henniker, 1949) of pure water surface are ascribed. Since the surface of water has a high surface energy, it is readily contaminated (Drost-Hansen, 1965). Even minutely contaminated surfaces display certain unusual elastic characteristics such as the Plateau-Marangoni-Gibbs effect (Scriven and Sternling, 1960). In contrast to surface tension, the role of elasticity in free-boundary flows has not been widely recognized (Scriven, 1946). Yet elasticity is to be generally expected, since even in the laboratory the likelihood of a clean surface is remote (Davies and Rideal, 1963). In a natural stream surface contamination is a certainty. Elastic properties of a contaminated surface can be attributed to local variations in surface tension (Whitaker, 1964) which exert a two-dimensional surface pressure, p, in the plane of the surface (Davies and Rideal, 1963) p=uo-cT

(7)

In Equation 7 uois the surface tension of a clean portion of the surface and u is the surface tension of a portion with adsorbed molecules, The surface pressure is always positive for water with adsorbed molecules. The forces resulting from these pressures are greater in regions of lower surface tension containing adsorbed molecules, Accordingly, the surface force is directed from a region of low surface tension to one of high surface tension. 58 Environmental Science and Technology

In the case of turbulent fluids the surface pressure and associated force have a definite effect on the process of interfacial renewal (Rukenshtein, 1965). Eddies continuously bring bulk fluid to the surface and thereby increase the local surface tension. Adjacent surfaces have lower surface tensions and there is therefore a tangential force tending to resist surface renewals (Davidson and Cullen, 1959). The importance of the gradients rather than the absolute value of surface tension has been demonstrated experimentally (Davies and Kahn, 1965). Surface elasticity is apparently the most important property affecting the renewal of the surface (Berg and Acrivos, 1965). An index of surface elasticity can be obtained by measuring the surface pressure, p , in conjunction with surface concentrations to produce force-area curves (Davies and Rideal, 1963). From these curves is obtained a two-dimensional compressional modulus (dynes per centimeter) relative to a clean surface. The value of the modulus depends o n the state of the surface film and generally increases as surface tension decreases (Bikerman, 1958). The surface modulus as usually measured would be zero for the standard of a clean surface. However, even a clean surface of water should have an absolute resistance to surface compression due to its high elastic strength. Thus there should be a positive value for the two-dimensional compressional modulus for any water surface. For this discussion such a modulus is termed M , and is used to reflect the surface elasticity of a water surface. An eddy in the vicinity of the surface may approach the surface with a certain dynamic thrust which could cause a dynamic pressure proportional to the product of the fluid density, p , and the square of the eddy velocity, c. It is suggested herein that the resistance to this dynamic pressure comes from the elastic character of the surface region. If the compressional modulus reflects this character, the resisting pressure due to compression of a surface region of depth, x , would be M,/x. The resistance must equal the dynamic pressure for a condition of surface stability; thus MS pv2 =

; -

An eddy, or portion of an eddy, traveling through the surface region of depth x, with velocity u, would have a residence time in the region proportional to xjo. The renewal rate, r, could then be proportional to uix. Using Equation 8 we have: (9) Levich (1962) presented a n equation for gas transfer containing which can be viewed as the case of Equation 6 with the coth term equal to unity (intense mixing). The remaining terms in his equation are equivalent to the surface renewal, r, and equal pr3/u, where u is the surface tension. When this expression is compared with Equation 9, the similarity is noted, except that M , appears in the latter equation while u is contained in Levich’s. The eddy velocity may be proportional to the Kolomogoroff velocity factor, (UE,)~’?, in which v is the kinematic viscosity and Es is the rate of dissipation of energy per unit mass of water near the surface. This proportionality, L; = C3(~Es)l/4, substituted in Equation 9 yields

The energy dissipated in the vicinity of the surface, E,, should be proportional to E, the energy dissipated in the fluid as a whole. If E, equals C2E,the resulting expression for r is ClCz314C33pY3i4E3’4 r = (11) Ms The form of Equation 11 is similar to a n equation previously proposed (Dobbins, 1964), with one difference. The latter equation contains surface tension as a parameter where Equation 11 has the modulus of compression, Ms. The significance of this difference is that values of M , increase in the presence of surface-active materials, while surface tension values decrease. Thus Equation 11 is in qualitative agreement with reported observations that surface-active materials retard surface renewals. Quantitative evaluation of Ms in turbulent systems could lead to the use of Equation 6 for predicting the effects of surface-active materials o n the absorption coefficient. This latter aspect is the subject of continuing research; however, for the work reported herein with distilled water, M,>is essentially constant. It has been suggested (Dobbins, 1964) that the effective film thickness, L: might be related to the eddy size L. =

c q

(2)

e-1

1

iELPY

II

. “ l L I

I

r U N I YUL

I*

I

v3_v E

CYLIN~ER

Figure 1. Experimental system

1’:

where C 4 is a constant which should be determined by the system. The depth of surface region, x, as used in Equation 8, is not the same as the effective film thickness suggested in Equation 12. The former value represents the depth where resistance to surface overturn is extant. Indeed, this depth could be much less than the size of an eddy and could retard motion as a portion of a n eddy moves through this region. If x were assumed to represent the film thickness, it would be inversely proportional to vl12Elz. Such an assumption would assign v 1 I 2 to the denominator rather than v ~as’in ~the numerator of Equation 12. In addition, the energy, E, would be raised to the 1/2 power instead of the power. These changes would not permit the consistency in the experimental results demonstrated in this work. The following relationship follows from Equations 11 and 12. The number of constants is reduced to one, Cj.

This equation is independent of the energy dissipated and thereby provides a key relationship for experimental verification of the proposed equations. Equation 13 predicts that rL8 should be constant at a given temperature and that its temperature dependency is due mainly to the terms in the numerator. Experiment Apparatus and Procedure. The apparatus, shown diagrammatically in Figure 1, was designed to allow determination of the amount of gas absorbed in a liquid by measuring the decrease in volume of the gas phase under constant pressure. A tube connects the absorption cylinder to a mercury manometer which contains an electrode to close a relay circuit when the mercury reaches a level of 763 mm. F o r a lower pressure, caused by absorption during operation, the circuit is broken

and the solenoid valve opens to allow water to be injected from the buret under auxiliary pressure into the pressurecontrol cylinder. The gas in the pressure-control cylinder is in free communication with the gas in the absorption cylinder and is compressed until the pressure returns to 763 mm. At this pressure the mercury comes in contact with the electrode and causes the solenoid valve to close. The increments of water added are as small as 0.05 cm. and equal the volume of gas absorbed during a very brief interval. The total volume of gas absorbed to any time is given by the reading of the buret. These readings are made at the completion of each injection and associated times recorded to 0.01 minute. Figure 2 shows the absorption cylinder, which is a Lucite cylinder with flanged ends and removable top and bottom. A thermometer extends into the upper portion of the cylinder, which contains gas during the operating phase. The absorption unit is provided with a lattice mixer driven vertically in simple harmonic motion with a stroke of 2.5 cm. by a variablespeed motor. A rocker arm extends through the side of the absorption cylinder through a completely sealed fitting and connects to the drive mechanism. Mixer speeds may be set in the range of 3 to 200 oscillations per minute. The term “cover” used in this work refers to the depth of water above the upper mixer blade at the top of a stroke during an experiment. The absorption cylinder, manometer, and pressure-control cylinder are set in water baths which provide controlled temperatures. The temperature in the absorption cylinder is recorded to 0.01 C., with a variation during a run not usually exceeding 0.10” C. Corrections to the data are made for the volumetric changes in the enclosed gas caused by temperature variations during a run. Before an experiment the entire system is filled with the gas to be used in the experiment. The gas is then forced from the system by admitting gas-free water, which is allowed to remain O

Volume 1, Number 1, January 1967 59

I ys

I

TIME, t

Figure 3. Course of absorption during experiment

TO M O T O R 3RIVE

8 R O W S OF S L A D E S

-

4

I1

PuBBEQ BELLOtVS 'YROJCH CYLINDER *ALL

I I

7"

5z

I

Figure 2 . Absorption cylinder until the desired temperature is reached, Gas is then introduced as the water is withdrawn from the cylinders to the desired operating levels. Distilled water was used for the majority of the experiments, A limited number of runs were made with tap water; however, no difference was noted and the results are presented together. The course of absorption during a typical experimental run is shown in Figure 3 by curve a-6-c-d. The dashed axis is the reference for the liquid with no dissolved gas. Y , represents the volume of gas absorbed at saturation. The solid axis shows the portion of the absorption curve used for computing the absorption coefficient. Point a indicates the possible dissolved gas in the liquid when it completely fills the absorption cylinder after repeated flushing of the system. During the interval from u to b, gas is introduced into the apparatus, and the water level and gas pressure are adjusted. Since the gas is in contact with the liquid, some absorption must take place. However, the amount of dissolved gas represented by point b is usually less than 10% of saturation. The buret reading is made at b and the mixer is started. The mixer is run for several minutes before an origin is established at c for computational purposes. It is in the region c to d that data are obtained for determining the absorption coefficient. A first-order equation is fitted to this region, using the slope method suggested by Thomas (1937) to obtain the reaction constant, K, and the value of Y indicated in Figure 3. The value of K thus obtained from the experiments is related to the apparent liquid film coefficient, K L ' , through the liquid volume, V , and the surface area, A,:

KL' =

KV

-

A" 60 Environmental Science and Technology

(14)

Results of Experiments. Results of experiments conducted in an apparatus similar to that used for the present work have been reported (Dobbins, 1964). The results for helium and nitrogen (Figures 4 and 5) were obtained under four different conditions of mixer stroke and cover. From these results, it was concluded that any mixing condition that results in a certain value of K L for nitrogen will also yield a definite value for helium, Thus, there is a consistent relationship between r and L which is independent of the mixing conditions which created them. This consistent relationship has been preserved in fitting the curves of Figures 4 and 5, although the curves represent a slight displacement from the original presentation. The results of a series of experiments in which absorption coefficients for helium were obtained at various temperatures in distilled and tap water are presented in Figure 6. Figure 6 a was prepared from a limited number of runs conducted under intense mixing conditions with the upper mixer blade breaking the surface at the top of each stroke. In Figure 6 b. c, and d, results are presented for mixer speeds of 100, 73, and 33 r.p.m., respectively, with a 1-cm. cover and stroke of 2.5 cm. These latter conditions were also used to obtain the results shown in Figure 7 for nitrogen, oxygen. and argon in distilled water. The data of Figures 6 and 7 are represented by straight lines on semilog paper, an observation reported by others (Elmore and West, 1961). The departure of oxygen and argon from the plotted straight lines of Figure 7 a and 6 , is due to their slight difference in diffusivity from that of nitrogen. Oxygen and argon values were not used in fitting the straight lines, but only to demonstrate that these other sparingly soluble gases followed the same temperature trend. Analysis of Experimental Results

Surface Areas. The data of Figures 4 and 5 for distilled water can be represented linearly in the lower speed region for any condition of stroke and cover. The departure from linearity in each case occurred when the liquid surface was noted to begin to ripple. A similar observation was made by Downing and Truesdale (1955), who reported that the absorption coefficient was linear with mixer speed until a speed was reached which caused vortices. When we assumed the departure from linearity of the curves in Figure 4 was due to an increase in area, ratios of the actual surface area to the horizontal projection of area were estimated by extrapolating the straight-line region and comparing this with the apparent values shown by the smooth curve. The

24

r -STROKE

COVER

1 - 1 . 2

I

1.0

I

w

1.4

E

b-

o

U

0 rn

0 4

a m

0.2

L i @0 40

r

80

L

120

160

280

240

200

M I X E R S P E E D lh ? E V O L U T I O N S P E R M I N U T E

Figure 4. Absorption coefficients for helium

16

STROKE COVER (VALUES IN C E N T I M E T E R S ) I 0 - 2 5 0

14-

1

7

I

__-

_1

Y

W

1

I

40

80

I23

,

16C

230

-___--240 23c

MIXER SPEED, IN REV3LUTIONS PER MINUTE

Figure 5 . Absorption coefficients for nitrogen ratios, designated by C,*, are plotted against mixer speed in Figure 8a for the three curves which departed from a straight line. This plot demonstrates concurrence with the results of Westerterp (1963), who reported that areas linearly depend on the agitation rate. The CAvalues obtained from each of the three curves have been plotted directly against the value of KL’ for helium in Figure 8b. When this is done, a single curve represents all the data. Similar results are obtained using the nitrogen data of Figure 5 . Thus a given value of KL‘ is always associated with a certain value of C‘.A, regardless of the mixing conditions which produced the value. It follows that there is a relationship between r and C.4 such that as energy increases in a surface region there is a maximum r value that can be attained with a smooth surface. Further increase of energy increases r. but only with an accompanying increase in area. Energy is envisioned as being absorbed by the compressional elasticity of a surface within its “elastic limit.” Increases of

energy imparted to a surface beyond this limit would cause marked deformations in addition to increased surface renewal frequency. Computation of r and L. The experiments were conducted so that at given speeds and temperatures mixing conditions were identical for the measurement of the absorption coefficients for helium and nitrogen. Thus, from Equation 6, the value of the surface renewal frequency, r, and the film thickness, L , at a given mixing condition, can be computed by inserting the known values of KL and D,,, for each gas and solving the resulting two equations simultaneously. For helium and nitrogen, represented by subscripts 1 and 2 , respectively, the following equations follow from Equation 6.

Volume 1, Number 1, Januar) 1967 61

_

~

W

3.0

Te,m",

3 i-

C.

Nitrogen

10 15 20 25 30 35

1.40 1.63 1.88 2.16 2.46 2.77

5 w

0

2.5 0.65 0.60

VI z

z 5 w w

w

u

z -

W

0

_

Table I. Diffusivities of Gases in Water" (Values in sq. cm./sec. X 105)

070

360 0 65 0 so

Helium 4.09 4.76 5.49 6.30 7.16 8.08

Oxygen 1.59 1.85 2.14 2.45 2.79 3.14

Selection of values presented in detail elseuhere (Metzger, 1965). 25°C. values for nitrogen and oxygen accepted as a mean of reported values. Helium value at 25°C. determined by using a ratio of helium to nitrogen of 2.92, which was obtained experimentally and is in accord with reported values. Diffusivities at other temperatures computed using Stokes-Einstein equation.

050

I

0303

I

I

I

z

0200'

i I

7

A

15'

4

where

1

DISTILLED TAP

o e

25'

200

T C M P E R A T U R E , IN

DEGREES

30'

35O

CENTIGRADE

Figure 6. Absorption coefficients for helium

0

_Lo-----+

k o i "

In1 ' 3 0-

-EMPERATbRE,

rL2/Di

If the apparent values of the absorption coefficient are used (Kl' and K2'), Equations 16 and 17 are equally valid but yield apparent values (r' and L'). The apparent values can be related, using Equation 4, to true values as in: r' = Ca2r

L'

0 35

=

rdl 33 RPM

N DEGREES

RPll

N I T Q0GEN

CENTIGRADE

Figure 7. Absorption coefficients for nitrogen, oxygen, and argon 62 Environmental Science and Technology

=

(1 8)

L/C*

The solution of Equations 15, 16, and 17 depends upon a knowledge of the diffusivities of the two gases used. In addition, the variation of diffusivity with temperature is desired. Unfortunately, there is some lack of agreement in the literature concerning this information. Under these circumstances, some of the experimental results were interpreted (Metzger, 1965) in conjunction with reported data to make a reasonable estimate of the values required. The values used in this work are presented in Table I . Equations 16 and 17 were used to compute apparent and true film thicknesses and surface renewal frequencies using KL' values for helium and nitrogen from the smooth curves of Figures 4 and 5. The values computed with data from the linear portions of the curves are true values. Apparent values, computed from data which departed from linearity, were corrected to true values by using the C.4 values of Figure 8 6 and Equations 18 and 19. These computations form the basis for the plots of Figure 9, in which the value of r is plotted against L and r'(L')3. The data of Figures 6 and 7 were also collected under identical physical conditions in order to compute values of r and L. Results of computations for r'(L')3,L ' , and r' are plotted against temperature in Figure 10 n, b, and c, respectively, Apparent values have been presented because of the uncertainty of C, at temperatures other than 20" C. Verification of Proposed Equations. The range of computed film thicknesses shown in Figure 9 a compares with those reported by Bolin (1960), Kanwisher (1963), and Davis and Crandall (1930), who estimated a film thickness of 3.5 X 10 x 10-8, and 40 X 10-3 cm., respectively. These computed film thicknesses display a consistent relationship with the

~

Figure 8. Ratios of actual surface area to horizontal production of area from helium data

surface renewal frequencies which is in accord with the proposed equations. The values of r ’(L’)3 depicted in Figure 9 0 remain constant, until the water surface begins to ripple, then begin to decrease. This decrease is due to a n increase in surface area, since T ’ ( L ’ )is~ equal to rL3/cA.The coefficient, cA, may then be computed from these curves in addition to the manner previously discussed, When this is done, the same values of Cd4 are obtained as previously shown in Figure 8. Thus for a given temperature and surface modulus, rL3 remains constant in accord with Equation 13. Figure 10 u demonstrates that r’(L’)3 has a consistent decreasing trend with temperature for each series. The 20” C. values for the three series will all yield a n rL3 value of about 81 X loF6 cc. per minute when surface area corrections are applied. If the same C Afactors were assumed to hold at other temperatures, all the data would be represented by a single line. The slope of the lines (or single line) representing the three series is the same as for the group pv3 as predicted by the numerator of Equation 13. As for the denominator of Equation 13, M , may also be temperature-dependent. However, if its dependency is of the order of the variation in surface tension due to temperature, any change in M , would be masked by the larger change in v3. It would appear that the group C6]M8can be considered nearly constant over the range 1 5 ” to 30” C. and that M , is not strongly dependent on the system dynamics in distilled water. The validity of Equation 13 would therefore be established. The effect of temperature o n the film thickness is shown in Figure 10 h. These curves are in qualitative agreement with Equation 12, which predicts lower L’ values for higher energies and a decrease in L’ with the temperature for constant energy. Figure 10 c demonstrates the effect of temperature on r ‘ and is in qualitative agreement with Equation 11. Correlation with either Equation 11 o r 12 is not possible for the mixer data, since the value of E is unknown. However, these relationships can be used in Equation 6 to predict the effects of temperature on oxygen absorption. Effects of Temperature on Aeration. The effects of temperature on the absorption coefficient for oxygen is often rep-

-

CG

6

3s

y x 6 0 -

-

-~~

-

b‘ 03

--

-

5403”iu u E > *3-----

-

-

\-~ ~-

c5

5

13 -Rd;

0

53

IO3

? P - E O F i Llv‘ R C N E I Y A L , r > E ? I4 N U T E

Figure 9. Relation of r to L and r’(L’)3 computed from helium and nitrogen data resented by an equation of the form KT = NehT,where u and b are constants and T is temperature. The constant, b, may be obtained from the slope of a semilog plot of KL us. T . I t is common to relate temperature effects to a value at 20” C.; thus KT/KZ00= (eh)T-*20. When (eb)is replaced by 8, the equation takes the familiar form KT

=

K20(0)T--20

(20)

Equation 20 is not necessary when the film penetration model is used and, in fact, may be not entirely correct. It is presented to draw attention to the fact that the slope of each line of Figures 6 and 7 is related to 0. These slopes become steeper as the mixer speed decreases, indicating that the effects of temperature are more marked at the lower conditions of turbulence. There is no single value of O which can represent the entire spectrum of mixing conditions. Therefore, O would have to be a variable in Equation 20, constant values being applicable only over certain ranges of turbulence. Volume 1, Number 1, January 1967 63

Temp., “C. D,: sq. cm./sec. X 10; r per min. L , cm. X 103

d m coth C r KL, cm./min. Ratio

e

Table 11. Effects of Temperature on Absorption Coefficient for Oxygen Low Illtellslty High Intensitj Mlxlng Mixing 10 30 20 10 20 30 2 14 2.79 1.59 1.59 2.14 2.79 0 30 0.37 0.25 122 100 83 67 55 45.5 12.2 10.0 8.3 1.34 0 84 0.55 1.85 4.37 2.83 2.00 1.05 1.01 1 .OO 1.15 1 46 0.37 0 0196 0.0204 0.36 0.0188 0.34 0.0216 0.340 0.388 0 0287 0,0408 0.364 1 330 1 420 1.070 1.065 1 029 1 036 1 006 1.007

In addition to the variations in slope with mixer speed, the nitrogen slopes are flatter than the helium slopes for comparable mixer speeds and covers. This observation is in accord with the film penetration model, which predicts that the gas of higher diffusivity would be more affected by temperature changes. While the film penetration model will automatically account for this, the value of 8 in Equation 20 would have to be different for each gas, in addition to depending upon the turbulence. Table I1 has been prepared to illustrate how Equation 6 can account for temperature changes in oxygen absorption. Consistent values of r and L were selected for a condition of high and low mixing intensity. The effects of temperature were assumed to be due mainly to the v 3 i 4 term in Equations 11 and 12. The rate of dissipation of energy, E, was considered to be unaffected by temperature changes (Kozinski and King, 1966), which would seem appropriate for rivers. It is seen from the table that the film penetration model predicts an increase in KL for oxygen with temperature which depends on the turbulence and the temperature. The computed values of 8 range from 1.036 for very gentle mixing to 1.006 for rapid mixing characterized by distortion of the surface. Intermediate mixing conditions yield values between these limits and demonstrate a continuous variation in 8. The various values of 8 reported in the literature are in agreement with the latter observations. Gameson, Vandyke, and Ogden (1 958) conducted aeration experiments at weirs and reported a 8 value of 1.015. In a slowly stirred system, Downing and Truesdale (1955) reported 1.0212 and indicated lower values for higher stirring rates. Elmore and West (1 961) absorbed oxygen at low mixing intensities and observed a value of 1.0241, while in higher intensity experiments it was 1.0226. Kozinski and King (1966) reported data which result in a value for 8 of 1.036 for low surface renewal frequencies. The importance of recognizing that the liquid film coefficient, KL,is a more basic parameter than the reaeration coefficient, Kz, may be illustrated by the work of Elmore and West. They adjusted their experimental conditions to produce a Kz value which was nearly the same as found in Tennessee streams. The natural streams have depths of about 3 to 10 feet (Churchill et a[., 1962) while the mixer depth was about 1 foot. This means that the KL produced in the mixer was approximately one third to one tenth that found in the streams. To adjust the mixer to comparable stream conditions based on 64

Environmental Science and Technology

~~Ezzsz 6

1

1

I

I

I

I

100

90

80 1

I0

20 25 32 I5 TEMPERATURE, DEGREES C E h - l i ? A ? E

35

Figure 10. Relation of temperature to r ’ ( L ’ ) 3 .L’, and r’ computed from helium and nitrogen data

KL, a Kz value of three to ten times the stream value should have been used. At this higher mixing intensit] a lower value of 8 would probably be found and would be more appropriate for the natural streams considered.

Conclusions A rational physical model for the absorption of low-solubility gases into turbulent liquids is based on the concept that an interfacial liquid film is in a continuous state of random renewal. Proposed equations for the rate of renewal and thickness of the interfacial film were purportedly verified in laboratory experiments. When these equations are used in the basic model for gas absorption, it predicts effects of temperature on oxygen absorption which are in agreement with values reported in the literature.

Nomenclature

horizontal projection of surface area actual surface area ratio of A , to A,, concentration of dissolved gas in fluid bulk concentration of dissolved gas in liquid at initial time concentration of dissolved gas at saturation dissolved oxygen deficit molecular diffusivity molecular diffusivity of helium molecular diffusivity of nitrogen rate of dissipation of energy per unit mass for flow as a whole rate of dissipation of energy per unit mass in vicinity of surface depth first-order equation parameter (base e) per time interval liquid film absorption coefficient apparent liquid film absorption coefficient liquid film absorption coefficient for helium liquid film absorption coefficient for nitrogen, and reaeration constant (base e) liquid film thickness modulus of compression at surface exponent frequency of surface renewal time temperature in degrees Centigrade velocity volume of liquid aerated distance first-order equation parameter volume of gas dissolved at saturation kinematic viscosity ratio of rL2 to D , mass density constant surface tension

Sanit. Ena. Die. Ani. Soc. Cicil Engrs. 88, No. SA4, Proc. Paper 3199, 1 (July 1962). Danckwerts. P. V.. Ind. Ena. Chem. 43. 1460 (1951). Davidson. J. F., Cullen, E . J., Trans. Inst: Chem. Engrs. (London) 37, 122 (1959). Davies. J. T.. Kahn. W.. Chem. Enn. Sci. 20, 713 (1965). Davies; J. T., Rideal; E. K., “Interfacial Phenomena,” Academic Press, New York, 1963. Davis: D. S., Crandall, G. S., J . Am. Chem. Soc. 52, 3757 (1 930). Dobbins, W. E., J . Sanit. Eng. Dic. Am. SOC.C i d Engrs. 90, No. SA3, Proc. Paper 3949,53 (June 1964). Dobbins, W. E., “Mechanism of Gas Absorption by Turbulent Liquids,” Proceedings of International Conference on Water Pollution, London, England, September 1962, Pergamon Press, New York, 1964. Dobbins, W. E., “Nature of the Oxygen Transfer Coefficient in Aeration Systems,” in “Biological Treatment of Sewage and Industrial Wastes,” Reinhold, New York, 1956. Downing, A. L., Truesdale, G. A., J . Appl. Chem. 5, 570 (1955). Drost-Hansen, W., Ind. Eng. Clzetn. 57, 18 (April 1965). Elmore, H. L., West, W. F., J . Sanit. Eng. Dic. Am. Soc. Cicil Engrs. 87, No. SA6, Proc. Paper 2997, 56 (November 1961). Gameson, A. L. H., Vandyke, K. G . , Ogden, C. G., Water Eng. 62,489 (November 1958). Henniker, J. C., Rec. Modern Phys. 21, 322 (1949). Higbie, R., Trans. Am. Inst. Chem. Engrs. 31, 365 (1935). Kanwisher, J., Deep-sea Res. 10, 195 (1963). Kozinski, A. A., King, C. J., J. Am. Inst. Chem. Engrs. 12, No. 1, 109 (1966). Levich, V. G., “Physiochemical Hydrodynamics,” PrenticeHall, Englewood Cliffs, N. J., 1962. Lewis, W. K., Whitman, W. G., Ind. Eng. Chem. 16, 1215 (1924). Metzger, I., “Effects of Fluid Properties on Stream Aeration,” Doctoral Thesis, New York University, 1965. O’Connor, D. J., Dobbins, W. E., Trans. Am. SOC.Cieil Engrs. 123, 641 (1956). Rukenshtein, E., Intern. Chem. Eng. 5, 88 (1965). Scriven, L. E., Chem. Eng. Sci. 12, 98 (1946). Scriven, L. E., Sternling, C. V., Nufure 187, No. 4733, 186 (1960). Thomas, H. A., Sewage Works J . 9,425 (1937). Westerterp, K. R., Chem. Eng. Sci. 18, No. 3, 157 (1963). Weyl, W. A., J . ColloidSci. 6, 389 (1951). Whitaker, S., Ind. Eng. Chem. Fundamentals 3, 132 (1964).

Literufure Cited Berg, J. C., Acrivos, A,. Chem. Eng. Sci. 20, 737 (1965). Bikerman, J. J . . “Surface Chemistry, Theory and Applications,” Academic Press, New York, 1958. Bolin, B., Tellus 12, No. 3, 274 (1960). Churchill, M. A,, Elmore, H. L., Buckingham, R. A., J.

Receiced for reciew October 18, 1966. Accepted December 22, 1966. Dicision of Water and Waste Chemistry, 152nd Meeting, ACS, New York, N . Y . , September 1966. Incestigation supported in part by a Public Health Sercice Fellowship 1-F3-WP26,053-01 from the Dicision of Water Supply and Pollution Control.

A, = A, = C, = CL = C, =

Thin-Layer Chromatographic Method for Estimation of Chlorophenols M. G. Zigler and W. F. Phillips, Residue Laboratory, Campbell Soup Co., Camden, N. J. A rapid, nonsophisticated analytical method is used for estimation of certain chlorophenols known to affect the flavor of water adversely at the parts per billion level. The twodirectional thin-layer chromatographic technique employs two supplementary reagents, aminoantipyrine and silver nitrate, both of which provide independent sensitivities of less than 1 p.p.b. The technique has been applied t o surface waters before and after treatment and should be useful t o investigators studying the effects of water treatments.

rapid, selective, and highly sensitive method was needed for determination of chlorophenols in water. This type of compound is known to affect the flavor of water adversely at parts per billion concentrations. A number of procedures and modifications for determining phenolics in water have appeared in the literature (Ettinger and Ruchhoft, 1948; Ettinger et al., 1951; Lyttan, et al., 1946; Murray, 1949; Rosenblatt et al., 1954; Schmauch and Grubb, 1954; Simard et al., 1951). The method developed is a two-directional thin-layer chromatographic procedure. Extraneous petroleum ether ex-

A

Volume 1, Number 1, January 1967 65